AI Working Group: Enhancing Informed Participation in the Jupiter DAO

Note:
This document is dynamic and will be continuously updated to reflect the progress and developments of the AI Working Group (AIWG). As we advance through the various stages of the project, new insights, updates, and modifications will be incorporated to ensure that this proposal remains accurate and relevant to our goals. Please consider this document as a work in progress that will evolve alongside the project.

Note on Upcoming Changes with LFG V2:

We are aware that the LFG process will soon evolve into LFG V2, impacting the focus of our working group. The role of Project Specialized AIs may change and will need to adapt accordingly. We will update this document as more details on LFG V2 emerge. We already have preliminary ideas to align our proposal with the new framework, ensuring the AIWG continues to contribute effectively to the Jupiter DAO’s goals.

Changelog
  • Added a note on the upcoming LFG V2 changes and its potential impact on the AIWG
  • Added a new sub-section titled “Mitigating Strategies Against Hallucination” in the Technical Explanation, detailing the AIWG’s approach to reducing AI hallucinations.
  • Added references to the engagement gaps in the middle of the funnel in the Abstract and Introduction sections, emphasizing the AIWG’s role in addressing this challenge alongside information overload and uninformed voting.
  • Added small demo of what the prototype can help with. Including GIF
  • Included Data Decider LLM in the Roadmap under Phase 3
  • Added a Jargon Look-up Table
  • Changed some sentences, to make it more clear what the solutions do.
  • Added the Live Prototype Demo by Kash and Meow section
  • Moved the Technical Explanation Section to Notion Site (Character Limit)
  • Added JupCatdets’ tweet thread in Live Prototype Demo Section.
  • Added ImageAnalyzer feature to Phase 1 prototype description.
  • Updated LLM section to reflect plan for GPT-4o-mini in Beta phase and future use of LLaMA 3.1 70B or similar models.
  • Added SpaceStation and Jupiter Mobile App integration plans for Beta Jupiter Horizon AI in Phase 2 and Phase 3 objectives.
  • Added $200 funding for ImageAnalyzer and clarified return of unused funds if the prototype is replaced or discontinued.
  • Added Achievement of reaching 400+ Chats
  • Added a new Problem + Solution section addressing language barriers and the AI’s multilingual capabilities.
  • Added a reference to the Technical Explanation in the AI hallucinations note.
  • Adjusted Roadmap and Budget, to not include Prototype grant.

Abstract:

The Artificial Intelligence Working Group (AIWG) is dedicated to advancing the governance capabilities of the Jupiter DAO by introducing sophisticated AI-driven solutions. Our primary objective is to develop and deploy the Jupiter Horizon AI and Project Specialized AIs—tools that assist DAO/community members by providing relevant information, answering questions, and offering guidance on navigating the Jupiter platform and its projects.
By addressing current challenges such as engagement gaps in the middle of the funnel, information overload, and uninformed voting, the AIWG seeks to enhance transparency, increase community engagement, and build long-term trust within the ecosystem.
This proposal outlines our strategy for achieving these goals, starting with the creation of a prototype and progressing towards full integration within the Jupiter platform.
The AIWG invites the DAO to support this initiative, explore, and provide feedback on the prototype, as we strive to elevate the effectiveness of decentralized governance.


Introduction

The Artificial Intelligence Working Group (AIWG) is poised to play a transformative role in enhancing participation within the Jupiter DAO. By developing advanced AI-driven solutions such as the Jupiter Horizon AI and Project Specialized AIs, the AIWG aims to address current challenges in the DAO’s decision-making process, including engagement gaps in the middle of the funnel, information overload, and uninformed voting. These initiatives align with the broader goals of the Jupiter DAO, which seeks to create a more transparent, informed, and engaged community.

The inspiration for the AIWG stems from my own journey within the Jupiter ecosystem. As I explored the various projects and navigated the complexities of the DAO, I personally experienced the challenges of information overload and the difficulty of getting properly informed. As an avid ChatGPT user, I realized how beneficial it would have been to have a chat-based tool that could streamline the process of learning about the projects and making informed decisions. This experience led me to envision a working group that could develop AI solutions to make participation in the DAO more accessible and effective for everyone.

The AIWG is vital for Jupiter DAO, improving the quality of participation in voting and decision-making. By offering precise, timely insights, AIWG empowers the community to make informed decisions, fostering a more resilient decentralized organization.


Live Prototype Demo by Kash and Meow: Jupiter Horizon in Action

We’re thrilled to announce that during a recent Planetary Call, Kash and Meow gave a live demo of the Jupiter Horizon prototype, and the results were impressive! They explored its real-time ability to navigate the Jupiter ecosystem and provide accurate answers to complex questions. Both Kash and Meow were amazed at how well it performed.

This can be further seen in the recent tweet thread by the JupCatdets, where they showcased several cool ways to interact with the Jupiter Horizon AI, guiding users on how to maximize its utility.


Problem Statement

The effectiveness of the Jupiter DAO relies heavily on informed and active participation from its members. However, several key challenges currently hinder the decision-making process and overall engagement within the DAO:

  1. Engagement Gaps in the Middle of the Funnel:
    • While Jupiter excels at attracting new users (“top of the funnel”), converting these newcomers into active, informed, and engaged DAO participants (“middle of the funnel”) remains a significant challenge. Many users express initial interest but struggle to transition into more involved roles within the community (Catdet&CoC). This issue has been identified by the CWG (as noted by Slorg and Kemosabe) and highlighted on the August 29th planetary call. The challenge revolves around helping newcomers understand and explore the Jupiter DAO and the broader Jupiter ecosystem, thereby increasing their engagement and participation.
  2. Language Barriers Limiting Participation
    • As the Jupiter DAO continues to expand its global community, language barriers pose a significant challenge to participation and engagement. Many community members, especially those less proficient in English, may struggle to fully understand proposals, governance decisions, or even general platform documentation. This limitation can result in reduced participation from non-English speakers, hindering their ability to contribute meaningfully to the DAO and creating an uneven playing field for informed decision-making.
  3. Information Overload:
    • Members are often overwhelmed by the sheer volume of information distributed across multiple channels. This can make it difficult to focus on relevant data, leading to potential confusion and less-informed decisions.
  4. Limited Access to Tailored Insights:
    • Existing project documentation often lacks personalization, leaving members without the specific insights they need to fully understand the proposals they are voting on. This gap in tailored information can result in decisions that do not fully consider all relevant factors.
  5. Misinformation and Rumors:
    • The decentralized nature of information flow can sometimes result in the spread of misinformation. Without access to reliable and verified information, members’ decisions might be swayed by unverified claims, impacting the trust and integrity of the voting process.

These challenges directly affect community trust and engagement. When members struggle to stay informed or feel overwhelmed by the voting process, they are less likely to participate actively. This, in turn, erodes confidence in the DAO’s ability to reflect the true will of its members and undermines the integrity of its governance processes.

By addressing these issues, the AI Working Group (AIWG) aims to empower the Jupiter community with the tools and insights needed to make more informed and impactful contributions, ultimately enhancing the DAO’s effectiveness and fostering a more engaged and trusting community.


Proposed Solution: AIWG

Overview of the Solution

The Artificial Intelligence Working Group (AIWG) is dedicated to enhancing informed participation within the Jupiter DAO by developing two core AI-driven solutions:

  1. Jupiter Horizon AI: The Jupiter Horizon AI assists DAO/community members by providing information, answering questions, and offering guidance on navigating and using the Jupiter platform. Users can ask about platform features, project statuses, governance decisions, or tools available, and get clear, actionable responses that help them engage more effectively. By centralizing and synthesizing relevant information—the good kind of centralization—Jupiter Horizon helps users easily access the data they need to make informed decisions. For those eager to experience its capabilities, the prototype is already available for testing.
    Curious about its capabilities? Ask it directly! Engage with the prototype for a firsthand experience of how it helps users navigate Jupiter.
    Jupiter_Horizon_Usage

  2. Project Specialized AIs: These tailored AIs are developed specifically for projects that are seeking to launch on the LFG Launchpad within the DAO. Each Project Specialized AI aggregates and analyzes data related to its respective project, offering members detailed, context-specific insights. These AIs are designed to answer project-specific questions, help members understand complex details, and ultimately empower them to vote with a deeper understanding of each proposal. This ensures that projects selected for launch have the informed backing of the community.

Addressing the Problem Statement

The AIWG directly addresses the challenges identified in the problem statement by leveraging AI technologies to improve the quality of information available to DAO members:

  • Addressing Engagement Gaps in the Middle of the Funnel: The AIWG is designed to help users, especially those new to crypto, go from zero to understanding what the Jupiter DAO is. By leveraging the Jupiter Horizon AI, we aim to provide a personalized, “1 on 1 onboarding” experience onto the Jupiter platform. This AI will deliver clear, relevant insights that guide users through the complexities of the Jupiter ecosystem, enabling them to transition from the middle of the funnel to the bottom of the funnel and become actively involved in the community.

Jupiter_Horizon_Onboarding

  • Combatting Information Overload: Both Jupiter Horizon and the Project Specialized AIs streamline the vast amount of information available, presenting only the most relevant and actionable insights to users. This reduces the burden on members to sift through excessive data, allowing them to focus on what truly matters for each decision.
  • Breaking Through the Language Barrier: To combat the language barrier, the Jupiter Horizon AI offers multilingual capabilities, allowing users to interact in their preferred language. Whilst it provides real-time translations and clarifications of key documents, it also helps users grasp the meaning and nuances. This enables non-English speakers to fully engage with the content, fostering a more inclusive and informed community.
  • Providing Tailored Insights: The AI solutions developed by the AIWG cater to the specific needs of each member. Whether it’s understanding the nuances of a particular project or getting a broad overview of the Jupiter platform, these AIs provide personalized, in-depth insights that help members make well-informed decisions. Additionally, the AI can break down and explain complex aspects of projects in clear, understandable language, making it easier for members to grasp difficult concepts.
  • Countering Misinformation and Rumors: By centralizing verified and accurate information, the AI tools help mitigate the spread of misinformation. Members will have access to fact-checked and reliable data, ensuring that their decisions are based on truth rather than unverified claims. This not only enhances the quality of voting outcomes but also restores and strengthens trust within the community.

Milestone Achievement: 400+ Chats!

:tada: The Jupiter Horizon prototype has already handled over 400+ chats, answering questions and guiding community members through the Jupiter ecosystem. It’s just the beginning of how this AI will transform decision-making within the DAO! :tada:

Screenshot 2024-09-23 at 10.37.08

Note on AI Hallucinations: It is important to acknowledge that while the AIWG is committed to providing accurate and reliable information, AI technologies can sometimes produce information that may be incorrect—a phenomenon known as “hallucination.” We are dedicated to minimizing these instances through continuous refinement and rigorous evaluation. For more details on the techniques we use to combat hallucinations, please refer to the Technical Explanation. However, users should remain aware of this inherent risk.

Long-Term Vision

The AIWG is not just a short-term initiative; it is integral to the long-term vision of the Jupiter DAO. As the DAO continues to grow and evolve, the need for informed, active participation will only become more critical. By embedding AI-driven tools into the fabric of the DAO’s governance process, the AIWG will ensure that the community remains engaged, informed, and empowered.

Over time, the AIWG envisions these tools becoming indispensable assets for the Jupiter community. As they evolve and improve, they will continue to enhance the overall effectiveness of the DAO, fostering a culture of trust, transparency, and informed decision-making. By addressing the challenges of today, the AIWG is laying the groundwork for a more resilient and adaptive Jupiter ecosystem in the future.


Roadmap

The AI Working Group (AIWG) has outlined its development into distinct phases, each designed to build upon the last, moving towards the full integration of AI-driven solutions that will enhance informed participation within the Jupiter DAO.

Phase 1: Prototype Development

The initial phase of the AIWG focused on creating the Jupiter Horizon prototype, demonstrating the potential of AI-driven insights within the Jupiter DAO. While the prototype is already available for use, it primarily serves to showcase the capabilities of centralized and synthesized information. Despite current limitations, including hallucinations, it is a significant step toward providing accurate, useful tools for the community. The goal is to build on this foundation to reduce these limitations in future versions. For those eager to explore its capabilities, the prototype is already available.

Additionally, the prototype now includes the ImageAnalyzer, a new feature that generates descriptions, identifies the main goal, and extracts visible text from images at scrape time. This makes the images accessible to the bot for better interaction and engagement with content. Currently this is only analyzing images from JupResearch posts.

Phase 2: Grant

Second Phase: Grant

The second phase is focused on securing the necessary funding for the Beta Jupiter Horizon AI, which will further expand the AI’s capabilities and integrate it more deeply into the Jupiter ecosystem.

  • Beta Jupiter Horizon AI:
    • Objective: Secure funding to develop a more advanced beta version of the Jupiter Horizon AI, which will include a new architecture and integration into Discord, making it more accessible to DAO/community members.
    • Deliverables:
      • Development of a new architecture that improves the AI’s accuracy and reduces the likelihood of hallucinations.
      • Integration into Discord, providing members with a more user-friendly and accessible platform.
      • A collaborative effort, expanding the development team to accelerate progress and enhance the AI’s capabilities.
      • If possible, we aim to integrate the Beta version into the SpaceStation, providing easy access to the model for all Jupiter users, not just Discord members.

Phase 3: Trial Work Group Development

Once the necessary grants have been secured and the AIWG has proven its value to the DAO, and as we progress further in the working group system, the focus will shift towards evolving into a Trial Work Group. This phase will also mark the beginning of the development of the Project Specialized AIs, alongside significant improvements and scaling up operations from the Beta Jupiter Horizon AI.

  • Objectives:
    • Establish a core team to oversee the development and deployment of both Jupiter Horizon and Project Specialized AIs.
    • Begin the development of Project Specialized AIs for projects seeking to launch on the LFG Launchpad.
    • Introduce the Data Decider LLM: A custom LLM to filter out irrelevant or noisy information, ensuring that only the most relevant, high-quality data is used by the AI. This will greatly improve the quality and precision of the insights provided to users.
    • Refine and enhance the AI models based on feedback and testing results from the beta phase, leading to significant upgrades from the Beta Jupiter Horizon AI.
    • Develop a comprehensive plan for full integration into the Jupiter platform.
    • If possible, extend the integration of the Jupiter Horizon AI to the Jupiter Mobile App, providing seamless access to AI-driven insights for all users, whether on desktop or mobile.

Phase 4: Full Integration and Continuous Improvement

In this final phase, the AIWG aims to fully integrate its AI models within the Jupiter platform and website and continue improving these tools to adapt to the evolving needs of the DAO. This phase will also focus on the AIWG’s transition into a permanent Work Group, ensuring long-term sustainability and impact.

  • Objectives:
    • Integrate Jupiter Horizon and Project Specialized AIs into the Jupiter website, making these tools universally accessible to all community members.
    • Establish ongoing processes for continuous feedback collection and model refinement to ensure that the AI tools remain relevant and effective.
    • Expand the scope and capabilities of the AI models to include new features and technologies as the DAO grows.

Over the past four months of building in silence, the project has evolved significantly based on new insights and developments. Therefore, while the intention is to achieve full integration and continuous improvement, the specifics of this phase—and potentially Phase 3—are likely to evolve as we move forward. This evolution will be driven by continuous community feedback to ensure that the scope aligns with the community’s expectations and priorities.


Budget and Resource Allocation

The AI Working Group (AIWG) is structured to evolve gradually, with resource allocation planned for each phase of the project. While we’ll focus on the immediate funding needed for the Beta Jupiter Horizon AI, we’ll also outline the resource needs for subsequent phases, keeping in mind that these will be adjusted as the project progresses.

Phase 1: Prototype Development

The initial phase of the AIWG involved creating the Jupiter Horizon prototype, which served as both a proof of concept and a helpful tool for the DAO, showcasing the potential of AI in improving decision-making and engagement. The prototype has provided valuable insights and functionality, but we do not require funding for it. Instead, we are now shifting focus towards developing the Beta Jupiter Horizon AI, which will build upon the success of the prototype.

Phase 2: Beta Jupiter Horizon AI

This phase focuses on refining and expanding the capabilities of the Jupiter Horizon AI, with a particular emphasis on improving its accuracy and integrating it more deeply into the Jupiter ecosystem.

  • Resource Needs:
    • Developer Expansion: To scale up the capabilities of the Jupiter Horizon AI, we will need to expand the development team.
    • Advanced Data Sources: Continued investment in APIs and other data sources will be essential to improve the accuracy and depth of the AI’s insights.
    • AI Token Usage Costs: Budgeting for the token usage required to run the AI, including API calls and other operational expenditures.
    • Database Costs: Implementation and maintenance of vector databases to store and retrieve embeddings efficiently.
    • Server Costs: Funds to support running the AI code on a reliable server 24/7, ensuring the AI is always accessible to DAO/community members.
  • Funding Note: The exact funding required for this phase will be determined based on the outcomes of the prototype phase and evolving project needs.

Phase 3: Trial Work Group Development

As the AIWG secures its position as a Trial Work Group, resource allocation will shift to support the expanded scope of work. This phase will include significant enhancements to the Jupiter Horizon AI and the development of Project Specialized AIs. However, given that this phase is quite far into the future, the specifics of the resource allocation may change as the project evolves.

  • Resource Needs:
    • Core Team Establishment: Resources will be allocated to form a dedicated core team responsible for overseeing the development of both Jupiter Horizon and Project Specialized AIs.
    • AI Hosting Costs: Budgeting for hosting services to ensure the AIs are reliably available.
    • Fine-Tuning Costs: If fine-tuning is decided upon, funds will be needed to cover the costs of this process.
    • Experiments and Innovation Costs: Resources allocated for ongoing experimentation with new techniques and technologies to keep the AI models at the cutting edge.
    • Operational Expansion: Resources will be required to support the expanded scope of operations as the AIWG transitions to a Trial Work Group, maintaining databases, servers, and other infrastructure.

Phase 4: Full Integration and Continuous Improvement

This phase is anticipated to focus on fully integrating the AI models within the Jupiter platform and website, as well as the AIWG’s transition into a permanent Work Group. However, this stage is quite far into the future, and the specifics are too uncertain to outline in detail at this time. This section will be updated as the project progresses and becomes more defined.


Team

Current Team Members

  • LiamVDB (Team Lead): I’m a community member of Jupiter (a Catdet) and the founder and lead developer of the AI Working Group (AIWG). I’m currently studying Computer Science, and over the past four months, I’ve been deeply involved in learning about AI. I’ve immersed myself in both the basics and the latest technologies to ensure that we’re building the best AI systems possible. My passion for AI and commitment to the Jupiter DAO have driven me to develop the Jupiter Horizon prototype, and I’m excited about what we can achieve together.
  • Milan (Recent Graduate in AI): Milan recently graduated with a degree in Civil Engineering, Computer Science, specializing in AI. He’s also eager in joining the work group and bringing his fresh perspective and specialized knowledge to help us refine and evolve our AI-driven solutions.

Future Team Expansion

We’re actively looking to expand the AIWG team. If you have experience in AI development, data science, or related fields and are passionate about contributing to the growth and success of the Jupiter DAO, we’d love to hear from you.


Technical Explanation [Moved]

This section dives deep into the technical aspects of the AI Working Group (AIWG), covering topics like model selection, LLMOps, Agentic RAG, AI refinement strategies, and the evolution of models across the different phases of our roadmap. It’s intended for those with a strong interest in AI development or potential team members looking to join the AIWG.

To explore these details further, visit our dedicated Notion page:

AI Working Group - Technical Explanation


Conclusion

Engage and Contribute

The AI Working Group (AIWG) is poised to bring transformative change to the Jupiter DAO. By harnessing advanced AI technologies and frameworks like Agentic RAG, we aim to elevate the decision-making process, making it more informed, transparent, and accessible to all community members. This is more than just an initiative to build sophisticated AI models; it’s about enhancing the very foundation of how the DAO operates.

I encourage everyone to provide feedback on this proposal. Your insights will be invaluable as we refine our approach and ensure that the tools we develop truly meet the needs of the Jupiter community. Additionally, I invite you to use and share the prototype I’ve developed. It offers a glimpse into what’s possible and serves as a starting point for the innovations that lie ahead. If you find this proposal interesting or impactful, please consider sharing it with others in the community. Your engagement will help us better understand how to improve and expand its capabilities.

Future Vision

Looking ahead, the AIWG envisions a future where AI-driven insights become a fundamental part of the Jupiter DAO’s governance. As we progress through the phases outlined in our roadmap, our tools will play a crucial role in enhancing decision-making and fostering greater community trust and engagement. This initiative is not just about the technology; it’s about building a stronger, more cohesive DAO where every member has access to the information they need to make well-informed decisions.

Ending Note

Thank you for dedicating your time to review this proposal. Over the past four months, significant time has been devoted to researching AI, preparing this document, and developing the prototype. The journey has been nothing short of incredible, expanding my knowledge and deepening my passion for AI and decentralized governance. I am excited about the possibilities that lie ahead and the potential impact this project could have on the Jupiter DAO.

The journey has been challenging, but the potential rewards are immense. With your feedback and engagement, we can take the next steps toward realizing this vision, ensuring that the Jupiter DAO remains at the forefront of innovation and excellence.

14 Likes

A strong and innovative proposal that tackles key challenges in the Jupiter DAO. The AI tools, especially Jupiter Horizon AI, promise to enhance informed decision-making and community engagement. Excited to see this in action!

2 Likes

Im gonna be critical here. How is this any different from me punching the voting prompt into chat GPT? like we are using AI to generate an analysis?

So if this issue is uniformed voting: my solution is this: why are people uninformed? do we need a TLDR for votes? are people not discussing pros and cons? do we need more media content,clips for the pros and cons of each vote?

Im gonna be honest, i apprecaite the post, but id defs vote no to this. Doesnt need a work group imo

3 Likes

Thanks for your feedback!
The Jupiter Horizon AI is specifically designed to go beyond what a generic AI like ChatGPT can offer. It’s not just about generating analysis—it’s about providing accurate, contextual answers to specific questions about projects within the Jupiter DAO, drawing from the most relevant data sources, like Discord announcements, Twitter posts, JupResearch posts, and the entire Jupiter SpaceStation. And this is just the beginning; as we progress through the roadmap, the AIs will only get better and better at addressing the unique challenges our DAO faces.

The problem statement in the proposal outlines the issues more thoroughly than the snippet you quoted, and the Proposed Solution section goes into detail about how the AIs will tackle these challenges head-on.

I also agree that TLDRs, media content, and discussions are crucial. The AIWG isn’t meant to replace these; rather, it’s an additional tool to help streamline and enhance the information available. The idea is to reduce the barriers to getting informed, not to bypass other important forms of engagement.

As for the workgroup, the goal is to ensure the AI evolves with the needs of the community, which is why we’re proposing a dedicated team. That said, I appreciate your perspective, and it’s important to have these discussions.
Thanks for being honest and for engaging with the proposal!

3 Likes

Yeah for sure. You’ve responded to me perfectly and I can’t fault you for that.

My argument is as follows:

  • we don’t need a work group for AI
  • We shouldn’t need to fund this, if someone wants to summarise they can do it on their own good will
  • Using AI to summarise announcements kinda replaces the need for the team to post announcements in the first place. Why would they even bother?
  • These work group proposals are getting a little out of scope of JUP’s core values and needs- I’ve heard people propose JUP music group and my question is “why the fk would defi exchange need that” lol.

No hate at all mate! In fact I’m proud you raised this idea and I commend you for you that.

I like to treat this like the show shark tank. And for that reason . I’m out

11 Likes

Thank you for your kind words! I’m thrilled to have your support and can’t wait to see the positive impact this project will have on the Jupiter DAO community.

3 Likes

Thank you for your thoughtful feedback and respectful tone—I genuinely appreciate the conversation we’re having. I understand your concerns, and I think there might be a slight misunderstanding about the purpose and scope of the AI Working Group.

First, the AIWG isn’t about replacing human effort or the need for team announcements; rather, it’s about enhancing and supporting the community’s ability to stay informed amidst an ever-growing stream of information.
There are people like you who are very informed, making it easier to stay updated. However, this solution also targets the middle part of the funnel—the people who are interested in Jupiter but aren’t informed enough or don’t yet know enough to become a true Catdet. For them, the entire Jupiter ecosystem can seem like a vast sea of information, very hard to sail. Staying up to date can also be quite time-consuming, and sometimes updates happen so fast that they can be missed.

For example, after the Jupiter Lock release, I heard something about it in a general discussion, but I couldn’t remember reading an announcement about it. So, I turned to my own prototype and asked it what the Jupiter Lock was, and the message it provided was quite helpful.
The AI tools we’re proposing are designed to complement, not replace, the communication efforts by the team and the community. They would serve as an additional resource to help members digest and contextualize information more effectively.

Regarding the scope of workgroups and the need for funding, I see where you’re coming from with the idea that someone could summarize information on their own goodwill. However, the AI we’re developing is designed to do much more than just summarizing announcements. It can help break down and explain complex concepts, like how the JLP (Jupiter Liquidity Pool) works—a topic that even I find challenging at times! The AI isn’t just a summarization tool; it’s a comprehensive resource that can provide detailed explanations, answer specific questions, and offer insights that might not be easily accessible otherwise. By funding this initiative, we can ensure that all community members, regardless of their technical expertise, have access to the information they need to participate fully and effectively in the DAO.

I respect your perspective and thank you for raising these points—it’s through discussions like these that we can refine our proposals and ensure they meet the needs of the community.

5 Likes

The proposal that you’ve outlined is detailed and ambitious but in my opinion it doesn’t clarify some major issues in a way that is understandable by everybody

  • List item How these improved versions of the chapGPT AIs will specifically address these issues in a measurable way, i.e. what would be the metrics to differentiate relevant information from the irrelevant information. This should be outlined in detail so that it gives much more confidence.

  • List item Also,there are a lot of complex technical jargons that common people like me can’t understand i.e. Agentic RAG, maybe if you can simplify and elaborate more as how this will be achieved it would be much more understandable.

  • List item There is a note on AI Hallucinations and if it is an inherent risk affecting DAO decisions there should be a robust plan on mitigating these risks and this should be outlined in detail.

  • The request of funds and the breakdown of it must be given to understand the exact Bill of Quantity items.

  • The integration of plans to the Jupiter platform must be furnished with detailed explanations as to how this will be achieved and what types of technical and operational challenges can be encountered and how these challenges will be dealt with.

3 Likes

Thank you so much for your detailed feedback. It’s incredibly constructive, and I really appreciate the time you’ve taken to engage with the proposal. I want to start by emphasizing that this document is dynamic and will evolve as we continue our journey with the AI Working Group (AIWG). We’re still very early in this process, currently focusing on securing grants, and will approach each phase step by step, constantly refining the document based on feedback and new insights. Your input is exactly the kind of feedback that will help shape this proposal moving forward, and I’ll be making adjustments accordingly.


This is a very relevant question. To differentiate relevant from irrelevant information, the AI will mostly rely on us to provide it with information from trusted data sources. While we’re currently focusing on trusted sources, we acknowledge that there may still be some noise or irrelevant information included. We already have plans for a “Data Decider” component to filter out such noise, but it isn’t a priority right now. This feature is part of our plan to be implemented during the Trial Working Group phase. I’ll make sure to elaborate on these plans in the proposal.

I completely understand that the use of technical jargon can be confusing. I’ve tried my best to avoid jargon throughout the proposal, and I’ve only used it in the “Technical Explanation” section, which is specifically intended for those with a strong interest in AI development or potential team members. To make it easier to understand, I’ll add a Jargon Lookup Table inside the Technical Explanation section where these terms will be explained in simpler language.

This is another very relevant concern. The “Agentic RAG” approach mentioned earlier is one of the strategies we plan to use to minimize hallucinations by ensuring the AI pulls from reliable, up-to-date data sources and relies on the data pulled from these. I’ll definitely be adding a section to the Technical Explanation that outlines our strategy for mitigating AI hallucinations in more detail. Thank you for pointing this out—it’s an important aspect that needs more visibility.

I understand that a clear breakdown of funds is critical, and it’s a big part of the proposal. At this stage, we haven’t outlined the budget for most of the phases in the roadmap yet, focusing instead on detailing the “Resources” we’ll need. As we progress through the roadmap, we’ll add specific budget estimates step by step. Right now, our main focus is on the first grant, for which we’ve provided exact figures. As we move forward, these figures will become clearer, and we’ll refine the budget accordingly.

The detailed integration of these AI tools into the Jupiter platform is something we plan to tackle as we move closer to the Trial Working Group phase. We won’t be working on this directly until we reach that stage, but we are mindful of the technical and operational challenges this integration may involve. As we approach that phase, we will provide a comprehensive explanation of how integration will be achieved, along with the challenges we anticipate and how we plan to address them.


I hope this helps clarify some of the concerns you’ve raised. Your feedback is invaluable, and I’m committed to making sure the AIWG proposal evolves in a way that addresses these important points. Thanks again for your input—it’s contributions like yours that will help us create a truly effective and valuable tool for the Jupiter community.

3 Likes

First of all these kind of things should not undergo a vote but should be decided directly by the team since its related to product/service development or some kind of add-on.

I know that today most of the people don’t read any proposal, nor care about them rather care just when is the vote, what they get for it and always blindly vote for yes. Summarizing more content won’t really change that those that don’t read or care become readers or be better informed. In my opinion content is not the problem, unless you prove me wrong :slight_smile:

Even so there could be other better uses for AI such as: walk-through website, help chat bot etc…

4 Likes

Reading through this AIWG group proposal has really caught my attention not only because I think it’s well presented but also it has real potential to not only make a difference on what it’s trying to solve but also directly or indirectly address a lot of things that other WGs are currently doing or could be be broaden to pull other WGs under it or used to cut the fat on other working groups hence my questions to @LiamVDB is:

How might the introduction of an AIWG in this Jup DAO streamline operations to the extent that it reduces or eliminates the need for certain existing working groups, and what are the potential implications of this shift?

4 Likes

Thank you for your feedback—I appreciate the insights you’ve shared. I’d like to clarify a few things regarding the AI and the process of bringing this proposal to a vote.

First, the AI we’re developing isn’t just a summarization tool. While summarizing information is one of its capabilities, the AI is designed to do much more. It provides contextual insights, answers complex questions, and helps users navigate the vast information within the Jupiter ecosystem. For example, it can break down intricate concepts, such as how specific protocols work or the implications of certain governance decisions, making this information more accessible to all members.

One of the key roles of this AI is to address the engagement gaps in the “middle of the funnel.” While Jupiter has been successful at attracting new users, converting these newcomers into active, informed, and engaged participants remains a challenge. The AI isn’t just about creating more content; it’s about making the existing content more actionable and easier to understand, especially for those who might not have the time or expertise to dive deep into every proposal. By offering a personalized onboarding experience, the AI helps bridge the gap between those who are deeply engaged and those who want to be but find it challenging due to the complexity or volume of information. This support helps users transition from simply being interested in Jupiter to becoming fully engaged members of the community, such as Catdets or Cats of Culture.

In essence, the AI serves as a tool to not only enhance the understanding of complex information but also to guide users through the Jupiter ecosystem in a way that encourages deeper involvement. It’s designed to make the journey from being a casual observer to an active participant more intuitive and less overwhelming.

As for why this should undergo a vote, I believe it’s more appropriate for this initiative to be funded and supported by the DAO because it directly impacts the entire community. The AI is intended to enhance how we all participate in governance, making it a community-driven tool. By bringing it to a vote, we ensure transparency and give everyone in the DAO a voice in deciding whether this is something we want to invest in collectively.

Regarding other potential uses for AI, like a website walk-through or a help chatbot, those are great ideas and certainly worth considering. However, these would require extensive work and are likely more suited for later phases of the project, such as the Trial Working Group phase or the full Working Group phase, where they would fall under the umbrella of full Jupiter platform Integration.

Thank you again for your input—it’s through discussions like these that we can refine and improve our initiatives.

1 Like

Thank you so much for recognizing the potential of the AIWG—I really appreciate your positive feedback! To clarify, the AIWG isn’t intended to replace other working groups but to complement and enhance their efforts by making information more accessible and improving overall engagement. While there might be some overlap, especially with the Web Working Group (which isn’t continuing), our focus is on using AI to support governance and participation. The goal is to empower and amplify the work being done by all WGs, not to replace them.

Out of curiosity, which working groups do you think the AIWG might overlap with or potentially replace? I’d love to hear your thoughts on this.

2 Likes

If you are suggetsing a discord style bot that repeats and summarises past announcements i feel that would be helpful.

for e.g. someone asked todat how much volume = what tier airdrop for jupuary.

I went and had a look and couldnt find it myself. So if there was a bot that automatically sent that info thatd be great.

Famous fox have a similair bot. is that what your referring to?

3 Likes

yes i agree with what youve said here Macximus. It skinda raises the question “is this worht our precious dao funding”

2 Likes

Not sure which WGs are specifically addressing these but in the areas of:

  • Governance such as proposal analysis & voting patterns
  • Community engagement such as sentiment analysis around proposals, votes etc…
  • Areas where we need marketing and communications such as audience targeting and content generation.
  • Research which can involve data analysis etc.
  • Human Resources such as performance monitoring for example analysing contributions & performance metrics within Jup DAO to support merit based rewards and also things like talent management which could involve identifying talent within community & make suggestion with this data to the Jup lead team to match skills to available roles.

Sure other WGs are already addressing some of these directly or indirectly but with the introduction of something like AIWG it can lead to more efficient operations and enhanced adaptability to changes within the Jup DAO ecosystem.

3 Likes

Hey Lochie,

I want to clarify something: the AI we’re developing is not just about repeating and summarizing past announcements. It does a lot more! The AI is designed to help with things like answering specific questions, guiding users through the Jupiter ecosystem, and even assisting newcomers in understanding how everything works—from zero to fully grasping what the Jupiter DAO is all about. It’s not just about summarizing; it’s about making information easier to find and understand for everyone, whether you’re new to crypto or a long-time member. I’m not sure why the focus keeps coming back to summarization, but this tool is much more versatile and powerful.

Regarding your example about the airdrop tiers, the Jupiter Horizon AI Prototype can already answer that kind of question. Here’s a screenshot from the prototype where it does just that.

It’s capable of pulling relevant information and delivering it quickly, so you don’t have to dig through past announcements.

Out of curiosity, have you used AI chatbots like ChatGPT before? If not, I’d encourage you to give our prototype a try to see firsthand what these tools are capable of. I think understanding how they work might help us better align our ideas and expectations.

Also, I checked out Famous Fox and even joined their Discord, but I didn’t find an AI bot like the one you’re describing. Could you clarify what you’re referring to?

Looking forward to your thoughts!

2 Likes

Thanks for sharing your thoughts. I understand where you’re coming from, but I’d argue that the core of the Jupiter DAO is to experiment and innovate. Our strength lies in trying out new ideas that have the potential to enhance our community and governance. We have sufficient funds in the DAO to support this kind of experimentation, and I genuinely believe that this AI initiative can bring meaningful improvements to how we access information, make decisions, and participate in the DAO.

By exploring and testing new tools like this AI, we’re not just spending for the sake of it—we’re investing in the future of the DAO through an innovative approach. If this works as intended, it could improve transparency, decision-making, and overall participation within Jupiter, making it well worth the investment.

2 Likes

Thanks for sharing your thoughts, Mbolorman—I appreciate the detailed breakdown. You’ve pointed out some really interesting areas where AI could potentially make an impact within the DAO, and I agree that there’s a lot of potential for AI to enhance operations in governance, community engagement, marketing, research, and even HR down the line.

However, most of these areas are not the current focus of the AIWG. Right now, we’re laser-focused on the roadmap we’ve outlined, which is centered around improving information accessibility, decision-making, and participation within the Jupiter ecosystem. While it’s exciting to think about how AI could eventually tackle some of the problems you’ve mentioned, those would likely be considerations for the far future as the project evolves.

While these ideas aren’t part of our current focus, I really appreciate you bringing them up. They could definitely inform our discussions as the AIWG grows and evolves.

For now, we’re committed to delivering on our immediate goals, but I’m definitely open to exploring these possibilities as the AIWG progresses. Thanks again for your input—it’s great to think ahead about where AI could take us!

2 Likes

Thanks for your response. I am impress with how you’ve given a lot of time & thoughts in addressing questions raised. That’s exemplary & shows your level of preparedness to deliver on your proposal. Hope we can integrate some of the ideas I suggested in the future considering the exciting possibilities of AI. It will be really helpful to make our DAO flourish even more.

4 Likes